LoRA for Frankie Foster from Foster's Home for Imaginary Friends
I trained it with 328 pics, which roughly 200 were screenshots and the rest fan art. I trained it at 20 epochs with a dim/alpha of 16/8.
LoRAs are quite time consuming so if you have any requests I'm open to commissions and donations on ko-fi.
Please play around with the LoRA weights. Let me know how it goes for you. Read stuff below on guidelines and known issues :)
Triggers:
The global activation tag is "FrankieFoster" it will generate her in a style that you want and should adapt well to other LoRAs and checkpoints.
The tag "FostersStyle" is there for you if you want her drawn in her actual cartoon style. It's less flexible this way because your restricting how it appears and stable-diffusion doesn't do well with cartoons. But most don't include stuff like this so I decided to do it.
To get her typical outfit: you may want to add, "green hoodie, purple skirt, ponytail, choker, midriff".
To get her black dress outfit everyone loves: "black dress, bare shoulders, jewelry, ponytail, bracelet, strapless, hoop earrings, black choker, high heels"
NB: can place in negative prompts if the clothes are stubbornly appearing.
Crucial Tips:
Due to the nature of this LoRA here is some advice on how I personally got decent results.
When generating the cartoon version of her by adding "FostersStyle" you may get pictures where she has many pupils in one eye. It seems common with stabke diffusion for cartoons but I found that saying "black pupil" can sometime help force it away. Though it can be easier to just edit the extras out with an image editor I guess.
When generating the cartoon version, consider not using the usual negative prompts. So don't put bad hands, bad fingers or even most negative prompt textual inversions. The reason is because cartoons are naturally weird (ie: Frankie in cartoon form is meant to have four fingers total and the tags may confuse the LoRA)
When generating the non-cartoon version of her, you may want to add "black eyes" to give her the color eyes she'd probably have.
Notes:
My experience has shown scale 0.7 ~ 1 works best; my best results were 1.
My prompt order shown in the sample pictures worked best for me.
Works on different models though weights may need to change for good results.
I used mainly the anyLora, Anything4.5 and Midnight Maple model.
Refer to my pictures for examples and what upscaler I used, keep in mind I have used other textual inversions to help with poses and negatives.
